Abstract
Laser Induced Breakdown Spectroscopy (LIBS) can be successfully used for cultural heritage applications, like pigment and stone identification or archaeological metal characterisation. In this work we focus on the study of wall paintings, working on a reference database of 30 commercial pigments traditionally used in murals and prepared with fresco, tempera (egg yolk, casein, animal glue) and oil techniques following ancient recipes. Two methods of signal analysis are tested for the automatic recognition of pigments and painting techniques. The first one is based on spectral lines identification for the detection of characteristic chemical elements. In the second one, we use two different chemometric models : soft independent modelling of class analogy (SIMCA) for the pigment identification and partial least-squares discriminant analysis (PLS-DA) for the technique characterisation. Results show the improvement of pigments and techniques recognition obtained with the multivariate analysis approach. As far as the real samples analysis is concerned, the signal processing has to be improved in order to remove non relevant and noisy information, such as background and thresholds that disturb the correct classification.
Keywords: Laser induced breakdown spectroscopy, Wall paintings, Pigments, Multivariate analysis
29